YOLOv8n-SPTS: Tiny Object Detection for Traffic Scenes (Looking for Feedback & ⭐ Stars)

[SonghanWu]

Hi everyone 👋
I’ve open-sourced YOLOv8n-SPTS, a lightweight YOLOv8-based model that focuses on small object detection in traffic scenes:

🔹 Repo: https://github.com/SonghanWu/yolov8n-SPTS

What’s inside?
A YOLOv8n variant enhanced for tiny objects (e.g., distant pedestrians, vehicles, traffic signs)
Detail-preserving convolutions and improved multi-scale fusion
An additional high-resolution detection head (stride 4) for small targets
Multiple attention mechanisms integrated in the backbone to better highlight important features

I’d really appreciate it if you could:
⭐ Star the repo if you find it useful
🧪 Try it on your own datasets and
💬 Share feedback, issues, or ideas for improvement

Thanks a lot for your time and support!

Model Demo:

[Rooster157]

I have starred yours please star mine: https://github.com/ArshiBansal/Stocks_Analysis & https://github.com/ArshiBansal/Cyber_Crime_Analysis

[SonghanWu]

Done, thank you!

[Rooster157]

Thank you